The National AI Compute Scorecard

Six major economies and multilateral blocs have established national AI compute initiatives, each adopting distinct funding mechanisms and metric definitions:

• United States: Public-private consortium coordinating donated commercial cloud credits and federal compute allocations.
• United Kingdom: Direct public procurement and capital expenditure for dedicated national supercomputers.
• European Union: Federally co-funded "AI Factories" integrated into the EuroHPC supercomputing network.
• India: Subsidized compute rental marketplace empaneled across commercial cloud service providers.
• France: Capital mobilization through private investment pledges and regional EuroHPC factory nodes.
• Japan: Targeted industrial subsidies to domestic cloud infrastructure operators and corporate balance sheets.

United States: National AI Research Resource (NAIRR)

The United States National AI Research Resource (NAIRR) pilot, launched in January 2024 by the National Science Foundation (NSF), coordinates private-sector compute contributions and federal lab access without direct government hardware acquisition (NSF NAIRR Overview).

• In-Kind Contributions: An estimated $100 million in private compute credits and software access from 28 nongovernmental partners, including NVIDIA ($30M in compute/software), Microsoft ($20M in Azure credits), OpenAI ($1M in model access), and Cerebras (up to 4 Exaflops capacity).
• Federal Participation: 14 participating agencies, with specialized access tracks such as NAIRR Secure (co-led by the Department of Energy and National Institutes of Health) for privacy-preserving medical and national security research.
• Reported Program Metrics: Over 840 research awards, 86 classroom initiatives, and support for more than 6,000 students across 50 states and territories (NAIRR Pilot Portal).

United Kingdom: AI Research Resource (AIRR)

The UK approach relies on direct public capital expenditure to procure and operate state-funded high-performance computing clusters (UKRI Announcement):

• Capital Allocation: £300 million committed in Phase 1 to construct and operate two primary supercomputing centers.
• Operational Hardware: Isambard-AI at the University of Bristol (equipped with 5,448 NVIDIA GH200 Grace Hopper superchips, funded via ~£225 million) and Dawn at the University of Cambridge (University of Bristol Release).
• Associated Infrastructure: An additional £750 million planned commitment for an exascale-class national supercomputing installation at the University of Edinburgh.

European Union: EuroHPC AI Factories

The European Commission operates a federated supercomputing network through the EuroHPC Joint Undertaking, combining €10 billion in multi-year public funding from the EU, member states, and partner nations (European Commission AI Factories):

Source: European Commission, EuroHPC.

The network encompasses 19 operational AI Factories, providing subsidized access to major supercomputers including Leonardo (Italy), MareNostrum (Spain), and LUMI (Finland), alongside an open call for up to 7 large-scale "AI Gigafactories."

India: Subsidized commercial cloud marketplace

India's compute provisioning operates under the IndiaAI Mission as an empanelment and subsidy model (IndiaAI Mission Analysis):

• Capacity Empanelment: 15 commercial Cloud Service Providers empaneled across four tender rounds, listing an aggregate registered capacity pool of over 45,000 GPUs.
• Price Discovery: Lowest-bid tender rates established benchmark pricing ranging from ₹29/hour (NVIDIA L4, monthly commitment) to ₹323/hour (NVIDIA B200 SXM, on-demand).
• Approved Volume: 237 projects approved and 93.18 lakh GPU hours sanctioned as of August 2026 (Lok Sabha Question 2807).

France: Investment summits and EuroHPC integration

France combines participation in EuroHPC infrastructure with international private capital mobilization:

• Summit Pledges: During the February 2025 AI Action Summit in Paris, announced multi-year international investment intentions totaled approximately €109 billion across sovereign and private infrastructure funds (Reuters).
• Public Systems: Operates the AI2F EuroHPC AI Factory connected to the Jean Zay supercomputer at GENCI, alongside planned commercial data center expansions.

Japan: Targeted enterprise subsidies and corporate balance sheets

Japan's strategy leverages direct grants to domestic cloud providers combined with overseas project investment:

• Ministry Subsidies: The Ministry of Economy, Trade and Industry (METI) allocated up to ¥72.5 billion in computational infrastructure subsidies under the Economic Security Promotion Act, including ¥50.1 billion awarded to Sakura Internet (METI Announcement).
• Private Financing: Japanese institutions and SoftBank provide syndicated project debt and corporate balance sheet financing for international compute frameworks.

Comparative summary of national compute models